Study On SOA-based Framework For Modeling And Simulation And The Technologies For Simulation Service Discovery

The development of technology has evolved system to system of systems. Modeling and Simulation (M&S) for these complex systems is focused on the coalition/joint development of models and simulation, and resources owned by different organizations or domains are reused and composed to meet the requirements of multidisciplinary cooperative simulation and multi-system joint simulation. The joint simulation for system of systems has demanded more M&S reusability and interoperability, and composable simulation has been the hot topics of M&S of complex systems. As a very important content, model composability has been one of the key research problems of composable simulation.Taking the Service-Oriented Computing (SOC) as the technology background, this dissertation has made an in-depth research on how to facilitate simulation model reuse and composition with Service-Oriented Architecture (SOA) and Web Service to promote the efficiency of composable simulation. From the views of SOA-based simulation approach and the discovery of simulation model services, the main content of the thesis are included: research of the SOA-based framework for modeling and simulation, the key technologies of simulation services description, mismatching, and selection, which are related to simulation service discovery; the design of the SOA-based simulation environment and the implementation of its prototype. The main work and innovations of this dissertation include the following parts:(1) The SOA-based framework for modeling and simulationFirst of all, the conceptual framework of SOA-based simulation is studied. SOA-based simulation and simulation service are explicitly defined, and the development process of SOA-based simulation and its advantage and challenge are discussed. After that, the high-level architecture of SOA-based simulation are explored, and a conceptual architecture is presented, which is Ontology-driven Open Simulation Architecture (OOSA), and the process model of SOA-based simulation, i.e. simulation service compositon, and some terms, the process view and technique framework are dicussed. Based on the above research, the SOA-based Open Modeling and Simulaiton Framework (SOMSF) is presented. The structure, functions and the related technologies in SOMSF are explored, and the key technologies involved in the thesis are explicated. SOMSF can provide overall support for technical, syntactic, semantic, pragmatic, and concept composability. (2) Simulation service discoveryUnder SOA, simulation service discovery is a prior precondition to reuse and dynamic composition of simulation models. This dissertation has separately studied three key technologies which closely related with service discovery: service description, service matching, and service selection.Firstly, the dissertation introduces existing Web service description models and the concept, express languages, and uses of ontology. After analyzing the semantics construct of simulation model, a two-level ontology-base description model for simulation service is proposed and the simulation service description ontology based on OWL-S/WSDL is build, which describes the technique, syntactic, semantic, and concept information of simulation service. Then the simulation domain ontology, concept, and simulation service description ontology are formalized. Finally, aiming to the deficiency of OWL-S to express dynamic behavior, a behavior ontology specification based on CADM (Core Architecture Data Model) is build, which can be used to support the expression and storage of behavior information.Secondly, based on the semantic description model for simulation service, a matchmaking model named"five types and three levels"is presented. Considering the difficult to construct the domain ontology and the matching efficiency, the type of Similarity Matchmaking, which is based on the semantics similarity degree, is stressed. Different similarity degree calculating formulas for description information of different level are established, and two key matching algorithms for semantic computing are studied. Finally, the performance of the matching method is compared with the existing web service discovery methods.Based on the characteristic of distributed simulation and the Quality of Service for Web Service, the Quality of Service for Simulation service (QoSS) is build and the aggregating rules based on graph is given, which can be used to guide the attribute calculating of composite QoSS. Furthermore, the QoSS-driven simulation service selection problem is modeled as a multi-objective combinatorial optimization problem, and the multi-objective plan model according to the characteristic of QoSS is build. Aiming to deferent user's requirements, the plan model can be solved by two different ways, and the algorithms based Particle Swarm Optimization (PSO) are provided with: Multi-objective and Multi-solution PSO (MMPSO) and Preference-Based PSO (PBPSO). MMPSO can achieve a group of Prato optimal solutions and PBPSO can only provide single solution. Finally, the performance of algorithms is analyzed and the applying situation of selection strategy is discussed.(3) Design and implementation of prototype Based on the above methods and technologies, combined with the fruits of our lab, a SOA-based Simulation environment - SOASim is designed, and the prototype and tools are implemented. SOASim can provide automated support for the design, development, publishing, deploying, discovery, composition and execution of simulation service. At last, an example is given to demonstrate the prototype system and partly validate the above approach and technologies.